The increase in persons with HIV (PWH) who receive diagnoses and are linked to care as a result of the Ending the HIV Epidemic in the U.S. (EHE) initiative¹ will further stress the HIV workforce, which remains in crisis, with a substantial proportion of providers dissatisfied with practice or planning to leave the workforce.^2-4 Achieving the goals of the National HIV/AIDS Strategy for the United States 2022-2025 (NHAS)⁵ will require increasing the number of infectious disease (ID) physicians, primary care physicians (internal medicine and family medicine), nurse practitioners (NPs), and physician assistants (PAs) who know how to select an ART regimen and recognize and respond to obstacles affecting adherence to ART and retention in care, such as mental health and substance use disorders, structural barriers to care, and internalized HIV-related stigma.^6,7 This will require increases in formal training opportunities and incentives to choose a career in HIV medicine, such as higher compensation for HIV care, reduced administrative burden, and assurance of safe and sustainable caseload.⁸ However, it is unclear how various provider types should be prioritized in these rebuilding efforts. A previous survey found that self-reported delivery of recommended HIV care was equivalent across provider types and that NPs were more likely than ID physicians to provide certain services considered key to providing comprehensive HIV care.⁸ But differences in key HIV outcomes among patients receiving care from providers of various professions and physician specialties has not been examined in the past two decades. We analyzed representative data from the CDC’s Medical Monitoring Project to evaluate differences in HIV outcomes among PWH by the type of providers they saw for HIV care, to inform efforts to strengthen the HIV workforce.

Table 1 displays characteristics and social determinants of health of PWH overall and by the type of providers they saw for HIV care. Compared with patients of ID physicians, patients of non-ID physicians differed by race/ethnicity and educational attainment and were more likely to report discrimination in the HIV care setting. Compared with patients of ID physicians, patients of NPs differed by race/ethnicity, age, poverty level, educational attainment, and health care coverage, and were more likely to have experienced unstable housing or homelessness or recent incarceration and to receive HIV care at a RWHAP-funded facility. They were less likely than patients of ID physicians to have had advanced HIV in the past 12 months. Compared with patients of ID physicians, patients of PAs differed by age, type of health insurance or coverage for antiretroviral medications, and poverty level, were more likely to have been unstably housed or incarcerated, to have used any injection or non-injection drugs in the past 12 months, and to receive HIV care at a RWHAP-funded facility. Compared with patients of ID physicians, patients of mixed providers differed by race/ethnicity, age, poverty level, and type of health insurance or coverage for antiretroviral medications, and were more likely to be unstably housed or incarcerated, to have unmet needs for ancillary services and advanced HIV in the past 12 months, major or other depression in the past two weeks, and to receive HIV care at a RWHAP-funded facility.

Table 2 displays the prevalence of HIV care outcomes during the past 12 months, overall and by the type of provider patients saw for HIV care. PDs assessing differences compared with patients of ID physicians were adjusted for patient characteristics, social determinants of health, and facility RWHAP funding status, according to the criteria described in Methods. There were no meaningful or statistically significant differences between patients of ID physicians and other provider types in ART prescription, sustained viral suppression, satisfaction with HIV care, or trust in providers. However, the prevalence of retention in care was higher among patients of non-ID physicians (aPD, 6.5; 95% CI: 2.7–10.3; P=0.001), NPs (aPD, 5.6; 95% CI: 2.3–8.9, P<0.001) and mixed providers (aPD, 12.7; 95% CI: 9.4–16.0; P<0.001). The prevalence of testing for syphilis, gonorrhea, and chlamydia among sexually active persons was higher among patients of non-ID physicians (aPD, 6.9; 95% CI: (0.8–13.1); P=0.026), NPs (aPD, 7.4; 95% CI: 1.6–13.3; P=0.012), and mixed providers (aPD, 13.5; 95% CI: 7.6–19.4; P<0.001). The mean antiretroviral adherence score was 1.8 points lower (95% CI: (−3.2–−0.4); P=0.011) for patients of NPs, on a scale of 0–100. Data for additional HIV outcomes are displayed in Appendix Table.

Compared with patients of ID physicians, higher percentages of patients of other provider types had characteristics and social determinants of health associated with poor health outcomes. After accounting for these differences, patients of non-ID physicians, NPs, and mixed providers were more likely to be retained in care and have recommended STI testing in the past 12 months, if sexually active. The mean antiretroviral adherence score was 1.8 points lower on a scale of 0-100 for NPs than ID physicians, which was statistically significant, but the magnitude was small and of unknown clinical significance. Other key HIV outcomes, including ART prescription, viral suppression, satisfaction with HIV care, and provider trust, did not differ meaningfully or statistically significantly by provider type.

Studies conducted from 1999-2005 found that more experience caring for patients with HIV, but not provider profession or physician specialty, was associated with delivery of recommended care and favorable clinical outcomes.^18-20 Our study is the first to examine clinical outcomes by provider type using a nationally representative sample of adults with HIV in the U.S.

Our finding that sexually active patients of non-ID physicians, NPs, and mixed providers were more likely than patients of ID physicians to have recommended annual STI testing,²¹ after accounting for patient and facility differences, is important, considering surging rates of STIs in the U.S., resulting in increasing health disparities.²² In addition to this finding, a previous MMP analysis found that NPs were twice as likely as ID physicians to report providing comprehensive sexual behavior-related risk-reduction services.⁸

There are several potential explanations for our finding that patients of ID physicians were less likely than others to be retained in care. It is possible patients of ID physicians on average require less frequent visits than patients of other provider types because of lower medical complexity due to having fewer comorbidities or differing social determinants of health for which we did not control. However, national guidelines recommend a viral load test at least every six months for all PWH.²³ All patients, regardless of level of complexity, who received care consistent with this recommendation would have been classified as retained in care (≥2 elements of HIV care ≥90 days apart in the past 12 months). Therefore, lower levels of complexity would not justify lower retention in care.

Differences in how care is delivered could explain higher retention in care and receipt of recommended STI testing among patients of non-ID physicians, NPs, and mixed providers than among patients of ID physicians. An MMP survey of HIV care providers found that equivalent percentages of ID and non-ID physicians, NPs, and PAs met HIV specialist criteria of the HIV Medicine Association or American Academy of HIV Medicine. However, one-third of ID physicians reported not providing primary care for HIV patients, defined as being the point of first contact, providing comprehensive care, and emphasizing prevention and coordination of care, compared with one in ten non-ID physicians and NPs and less than 1% of PAs.⁸ Another analysis of MMP data found that, among patients who needed HIV case management, those who received care from ID physicians were more likely than patients of NPs to have that need remain unmet, even after accounting for patient differences and the availability of onsite HIV case management services at the HIV care facility (publication pending), suggesting coordination of care may more often be lacking among some patients of ID physicians with complex needs. In addition, a 2014 MMP survey of HIV care providers found that NPs are more likely than ID physicians to provide care for >50 HIV patients and are five times as likely to be full-time HIV care providers,⁸ which aligns with previous findings that experience, rather than provider profession, is associated with favorable outcomes.^18-20

Retention in care and recommended STI testing are dependent on the coordination of care and prevention elements of primary care.²⁴ ART prescription and viral suppression, which did not differ by provider type, are more reflective of HIV disease-specific care than of primary care. Although 80% of ID physicians report having sufficient time with patients to provide necessary HIV information, only about 40% spend >20 minutes with HIV patients during follow-up visits, compared with two-thirds of NPs and PAs (8). A focused approach to HIV care and treatment by some ID physicians may exclude an emphasis on prevention and coordination of care that primary care physicians, NPs, and PAs routinely address.²⁵

Including providers with primary care training and experience in ID medicine practices could lessen differences in primary care outcomes. However, a potential barrier to broadening the HIV provider workforce is high levels of dissatisfaction with administrative burden, time available to provide necessary care, and remuneration among non-ID physician HIV care providers.⁸ Also of concern, the percentage of NPs planning to leave HIV practice within five years is triple that of ID physicians,⁸ which could have implications for meeting NHAS/EHE goals^1,5 and goals of the STI National Strategic Plan.²⁶

Other strategies for ensuring patients of ID physicians receive comprehensive primary care include ongoing training in primary care delivery, multidisciplinary case conferences, and use of decision support to nudge ID physicians to routinely provide recommended preventive care and assess their patients’ psychosocial needs.

Unlike patients of NPs, patients of PAs did not differ from patients of ID physicians in retention in care and receipt of recommended STI testing. Although the practice style and training of NPs and PAs has not been rigorously compared, one institution that trains both NPs and PAs noted that PA training uses the medical model, which focuses on treatment of the disease. NPs train on the nursing model, which focuses on treatment of the patient with the disease.²⁷ This patient-centered approach to treatment provided by NPs might account for the differences in our findings.

Our study had limitations. We recorded medical record information at the patient’s usual place of outpatient HIV care, including information from outside facilities, if available. It is possible some patients had tests performed elsewhere that were not available in the medical record of the HIV care provider,²⁸ resulting in lower estimates of retention in care and STI testing and potentially biasing our estimates. Regardless, our findings describe whether HIV care providers ensured and documented that their patients received care consistent with guidelines. Second, MMP does not collect information about informal physician consultations by NPs and PAs and its effect on outcomes. Our findings compare outcomes across models of care based on the profession of the provider of record. Some of these models of care might include informal consultation with other HIV specialists. Third, most data on social determinants of health and some data on clinical outcomes, e.g., substance use, depression, and anxiety, were self-reported and could be subject to information bias. Forth, data collection for a portion of the 2019 and the entire 2020 MMP cycle took place during the COVID-19 pandemic, which could have affected our findings. However, the overall proportion of persons linked to HIV care, prescribed antiretroviral therapy, and virally suppressed remained stable from 2019 through 2021.²⁹ Fifth, MMP data are cross-sectional; therefore, causality is not implied. Sixth, although our models accounted for differences in demographic and clinical characteristics and social determinants of health between patient groups, which increased our ability to identify independent associations between provider type and clinical outcomes, there could be residual confounding. Finally, response rates were suboptimal, but results were adjusted for non-response and post-stratified to known population totals by age, race/ethnicity, and sex from the NHSS using established, standard methodology. Even with suboptimal response rates, there is still value in results obtained from unbiased sampling methods.³⁰

Compared with patients of ID physicians, patients receiving HIV care from other provider types were more likely to have characteristics and social determinants of health associated with poor health outcomes and to receive HIV care at RWHAP-funded facilities. After accounting for these differences, most HIV clinical outcomes were equivalent across provider types. However, patients of non-ID physicians, NPs, and mixed providers were more likely to be retained in care and have recommended STI testing in the past 12 months, if sexually active. Strengthening delivery of comprehensive primary care by ID physicians who provide outpatient care for PWH and increasing the role of primary care providers in ID physician practices could help advance the national goal of improving clinical outcomes for PWH.